ADM Systems in the COVID-19 Pandemic: Poland

New report: 'Automated Decision-Making Systems in the COVID-19 Pandemic: A European Perspective' by AlgorithmWatch and Bertelsmann Stiftung – a special issue of the upcoming Automating Society Report 2020, to be published in October.

Country analysis: Poland

By Natalia Mileszyk and Alek Tarkowski

The techno-solutionist logic behind some actions of the Polish government in response to the pandemic is just more proof that ADM systems in Poland, if they exist, lack transparency, societal overview, and well-informed public discussion around the issue.

A mandatory home quarantine app, with face recognition

The Polish Ministry of Digital Affairs launched a Home Quarantine app on March 19 that uses GPS location, time-stamped photos, and face recognition to ensure that citizens stay at home. As per its terms of service, the government uses this app to ensure that people instructed to remain in quarantine do so.

The app routinely asks users to share their location which must match with their GPS location. They are also asked to take a photo at the location and complete a “task” within 20 minutes of receiving the message from the government. If this is not done, action could be taken against the person by the authorities.

From April 1, the app was made mandatory, which, in our opinion, is not proportionate (due to factors, such as, people sending images of themselves to government servers).

The Home Quarantine app most probably includes an ADM component, which uses automatic face recognition to confirm that the photos sent by users match the reference photo provided when creating an account. According to independent experts who have studied the application, the system most probably uses ADM components available as a component in the Azure cloud solution used by the app. No official documentation on this functionality (or lack thereorf) has been made available.

Bluetooth-based digital contact tracing

In June 2020, another application was launched – contact tracking ProteGO Safe app using Bluetooth technology to log connections between smartphones on a device and Exposure Notification protocol. The application evaluates the risk for COVID-19 exposure by using three colors: green – low, yellow – medium and red – high – colors are only recommendations to contact health authority. The application evaluates the risk based on following criteria: the length of the contact, the distance from the COVID-19 carrier, the date of the contact and certainty of the contact.

At the time of writing, ProteGO Safe did not distribute a single key allowing users to find out that people with a confirmed coronavirus diagnosis were in the vicinity. Over the past two weeks, not one of the 4,000 new COVID-19 carriers has sent information about their social contacts to the ProteGO Safe server.

In response to the development of new apps by the Polish government, Centum Cyfrowe Foundation has participated in the co-creation of “The Seven pillars of trust” – a set of important standards and rules [PL] to which these applications should adhere. These rules, in particular, recommend minimizing the data collected and having strict time periods for its retention, which states must follow in order to comply with fundamental rights.

ADM in the business sector

Some Polish businesses have implemented ADM systems to help fight the pandemic. Since 2012, Infermedica has been developing artificial intelligence tools for patient triage, and symptom checking, and these tools were adapted during the pandemic.

A curated COVID-19 pre-screening solution for patients, compliant with WHO guidelines, was deployed. The goal was to shift the burden of triage from health care practices, government-organized assessment centers, and emergency departments while helping all patients to quickly and accurately self-evaluate their risk of infection, and properly send them to the appropriate venue for care.

The COVID-19 screening protocol was first deployed by Symptomate, and then by other platforms, including Call Center Triage and Infermedica API.

Another example is FeverGuard, which is an AI-driven solution that combines analytic models with thermal image recognition to monitor body temperature and detect anomalies. By applying deep-learning, object tracking, and a temperature correction model, it can successfully extract human body temperature in real-time.

Download the full report as PDF (1 MB).

Automated Decision-Making Systems in the COVID-19 Pandemic: A European Perspective is a special issue of the report Automating Society 2020 by AlgorithmWatch and Bertelsmann Stiftung, to be published in October. Subscribe to the our newsletter to be alerted when the report is out.